Augmentation of human monocyte responses to LPS by the Protein S and Mer/Tyro3 receptor tyrosine kinase axis

Abstract

Resolution of the inflammatory response requires coordinated regulation of pro- and anti-inflammatory mediator production, together with clearance of recruited inflammatory cells. Many different receptors have been implicated in phagocytosis of apoptotic cells (efferocytosis), including Mer, a receptor tyrosine kinase (RTK) that can mediate recognition and subsequent internalisation of apoptotic cells. In this manuscript, we examine the expression and function of the Tyro3/Axl/Mer (TAM) family of receptors by human monocytes. We demonstrate that the Mer ligand, Protein S, binds to the surface of viable monocytes via phosphatidylserine (PtdSer)-dependent and -independent mechanisms. Importantly, we have identified a novel role for RTK signaling in the augmentation of monocyte cytokine release in response to LPS. We propose that low level PtdSer exposure on the plasma membrane of viable monocytes allows Protein S binding that leads to TAM-dependent augmentation of pro-inflammatory cytokine production. Our findings identify a potentially important role for TAM-mediated signaling during the initiation phase of inflammation.

Introduction

In response to injury or infection, dynamic temporal changes within the inflammatory microenvironment act to co-ordinate inflammatory cell recruitment(1), cell activation and tissue repair(2) and ultimately, regeneration(3). Apoptosis has a key role in the termination and resolution of the inflammatory response. Recruited inflammatory cells that do not emigrate from the inflammatory site(4), undergo apoptosis(5) and are subsequently cleared by phagocytosis (also termed efferocytosis)(6, 7). In addition, resolution of inflammation and the restoration of normal tissue function is facilitated by functional alterations in phagocytes as a consequence of signaling events downstream of apoptotic cell recognition (8, 9).

The Tyro3/Axl/Mer (TAM) family of receptor tyrosine kinases (RTK) are widely expressed on phagocytic cells in many different tissue settings and form one of the key molecular pathways by which phagocytes specifically recognise and subsequently internalise apoptotic cells(10). TAMs are able to bind to two specialized ligands, Protein S (Pros1) and Growth Arrest Specific protein 6 (Gas6)(11) that interact with the anionic phospholipid, phosphatidylserine, (PtdSer) exposed on the plasma membrane of apoptotic cells following loss of plasma membrane asymmetry (12, 13). Gas6 functions as a ligand for all three TAM receptors, whereas Pros1 serves as a ligand for only Tyro3 and Mer, with little or no binding to Axl(14). The presence of a highly conserved Gla-domain, containing multiple glutamic acid residues that are γ-carboxylated in a vitamin K-dependent reaction(15), confers the capacity of Pros1 and Gas6 to interact with PtdSer in a Ca²⁺-dependent manner(16).

TAM binding to Gas6 or Pros1 is critical for the physiological removal of apoptotic cells and the absence of TAM-dependent apoptotic cell clearance mechanisms is associated with failed spermatogenesis in the testis(17), photoreceptor degeneration in the eye(18) and development of features of autoimmune and autoinflammatory disease(17, 19, 20). Signaling down-stream of TAM ligation provides feedback inhibition of inflammatory cytokine production following Toll-like receptor (TLR)-driven responses(21, 22), allowing termination of inflammatory responses. Mer-dependent signaling in phagocytes promotes the internalisation of apoptotic cells via autophosphorylation at Y867 and activation of phosphatidylinositol 3- kinase (PI3K), phospholipase C (PLC)-γ2 and activation of Protein kinase C (PKC) leading to Focal Adhesion Kinase (FAK) phosphorylation that was not required for suppression of Nuclear factor κ light chain enhancer of activated B cells (NF-κB) activation(23).

In this manuscript, we have investigated the expression and function of TAMs by human monocytes, identifying a novel role for TAM signaling in the regulation of monocyte responses to LPS. We show that human monocytes express predominantly Mer, with low levels of Tyro-3 and little or no expression of Axl. Our findings confirm that viable monocytes expose low levels of PtdSer(24). Pros1 was shown to bind to the surface of viable monocytes via both a Ca²⁺-dependent and –independent mechanism, likely to represent PtdSer-dependent and TAM-dependent binding interactions respectively. Surprisingly, we found that Pros1 acts in a PtdSer-dependent manner to augment monocyte release of pro-inflammatory cytokines (TNF and IL-6) in response to lipopolysaccharide (LPS). Augmentation of monocyte pro-inflammatory cytokine release was attenuated when TAM RTK activity was blocked. We propose that low level exposure of PtdSer on viable monocytes allows Pros1 binding to monocytes that leads to augmentation of pro-inflammatory cytokine production, identifying a potentially important novel role for TAM-mediated signaling during the initiation phase of inflammation.

Materials and Methods

Reagents

Reagents were obtained from Sigma-Aldrich (Gillingham, UK) unless otherwise stated. Cell culture ware was purchased from either BDBiosciences (Oxford, UK) or Corning (NY, USA) and Iscove’s modification of Dulbecco’s modified Eagles medium (IMDM) (with the addition of 1% L-Glutamine, 25 mM HEPES) was purchased from Thermofisher (Paisley, UK). Human Pros1 was obtained from Enzyme Research Laboratories (Swansea, UK). Protein S lacking the Gla domain (ΔGla Pros1) was generated using overlap-extension mutagenesis to delete the Gla domain from the Pros1 gene using plasmid pCIS2M-PS(25). The following primers were used

A = 5’-TCGATTGCTAGCGCACAG-3’,

B = 5’-GAGCGAAGACAAACACGACGCTTCCTAACCAGG-3’,

C = 5’-CCTGGTTAGGAAGCGTCGTGTTTGTCTTCGCTC-3’,

D = 5’-GACACAAAGCTGAGCACACAT-3’.

These primers result in deletion of 135 nucleotides coding residues Ala1 to Leu45, leaving the pre-propeptide sequence (Met-41 to Arg-1) connected to Val46. The resulting PCR fragment was inserted, after digestion with NheI and BlpI, into the Pros1 gene in the pcDNA3.1+ expression plasmid to generate pcDNA3.1-ΔGla-PS. ΔGla-PS was expressed in HK293 cells and purified by immuno-affinity chromatography(26). Recombinant mouse Gas6 (full-length and Gla-less) was prepared as described(14) and provided by Prof. Greg Lemke and Dr. Erin Lew (Salk Institute, La Jolla, CA 92037). Phycoerythrin- or APC-conjugated mouse monoclonal antibodies (IgG1 unless otherwise indicated) against human CD14 (MΦP9, IgG2b), CD16 (3G8), CD62L (DREG56) CD162, (KPL1) CD11b (ICRF44) and isotype controls were obtained from BDBiosciences. Phycoerythrin (PE)-conjugated mouse monoclonal antibodies against human Mer (122518, IgG2b), Tyro3 (96201, IgG1) and Axl (108724, IgG1) together with relevant PE-conjugated isotype controls (IgG1 and IgG2b) were from R&D Systems (Abingdon, UK). A rabbit monoclonal anti-Mer for immunoblot analysis (clone D21F11) was obtained from Cell Signaling Technology (Danvers, MA). A second PE-conjugated mouse monoclonal antibody against human Mer (590H11G1E3, IgG1) was obtained from Biolegend, London, UK to validate flow cytometry findings. ELISA reagents and antibodies against IL-6 and TNF were obtained from R&D Systems. BMS777607 was obtained from Selleckchem (Stratech, Newmarket, UK). APC-conjugated Annexin V was obtained from Thermofisher and propidium iodide was from Biolegend (London, UK). LPS (E.coli 0127:B8) was purchased from Sigma (Gillingham, Dorset, UK). Recombinant Human TNF, Human TNF DuoSet ELISA and Human IL-6 Quantikine ELISA were from R&D systems (Abingdon, UK).

Blood cell isolation and culture

Peripheral blood leukocytes (human blood mononuclear cells and polymorphonuclear cells) were isolated from healthy volunteers as described(27). In brief, citrated blood (3.8%) was centrifuged at room temperature at 350g for 20 min, platelet-rich plasma was removed. Autologous serum was obtained by addition of CaCl₂ to a final concentration of 22mM. Leukocytes were separated from erythrocytes by dextran sedimentation and further fractionated using isotonic Percoll (GE Healthcare, Buckinghamshire, UK) 50%/70%/81% discontinuous gradients. Polymorphonuclear leukocytes (>95% CD16+ve, hereafter termed “neutrophils”) were harvested from the 70%/81% interface and cultured for 18h in IMDM supplemented with 5% autologous serum to induce apoptosis. Monocytes were enriched from the mononuclear cells (55%/70% interface) by negative selection using a Pan Monocyte Isolation Kit (Miltenyi Biotech, Bisley Surrey, UK) as described by the manufacturer. Monocytes were resuspended at 1.5x10⁶ cells/mL and cultured for 18h in IMDM supplemented with 5% autologous serum in Falcon tissue culture plates (BD Biosciences). Monocyte-derived macrophages (MDMs) were obtained by in vitro culture of monocytes for 6 days in IMDM supplemented with 5% autologous serum with the addition of 250 nM dexamethasone to induce a macrophage phenotype that predominantly utilizes Mer-mediated phagocytosis of apoptotic cells(28).

Assessment of phagocytosis of apoptotic cells

Phagocytosis of pHrodo-labeled apoptotic neutrophils by monocytes or MDM was assessed by flow cytometry as described(29, 30). To ensure TAM ligand-free conditions in the phagocytosis assay, cells were washed with HBSS without Ca²⁺/Mg²⁺ containing 2.5 mM EDTA to remove exogenous TAM ligands. monocytes were labeled with CellTrace™ Far red DDAOSE (Thermofisher) according to manufacturer’s instructions prior to phagocytosis assay. For experiments using the receptor tyrosine kinase inhibitor BMS777607, monocytes or MDM were pre-incubated for 20 min with 100 nM BMS77607 prior to co-incubation with pHrodo-labeled apoptotic neutrophils. For some assays, Pros1 or ΔGla Pros1 was added to a final concentration of 25 nM – as detailed in the Figure legends. MDMs or monocytes were then overlaid with apoptotic neutrophils at a phagocyte:target ratio 1:6. After co-culture for 40-45min at 37°C, non-ingested apoptotic cells were removed by aspiration and monocytes or MDM were detached by vigorous pipetting following incubation with 0.5% Trypsin/1mM EDTA solution for 5 min at 37°C. The percentage of monocytes or MDM that were fluorescent (corresponding to those that had ingested apoptotic neutrophils) was determined by flow cytometric analysis(31).

Labeling of proteins with fluorophores

Proteins were labeled with Dylight488 or Dylight650 dye as recommended by the manufacturer (Thermofisher). Following incubation with 7-10-fold molar excess of dye for 45 min in the dark at 20°C, the reaction was terminated by addition of glycine (final concentration 20mM). Labeled protein was layered onto a 7K molecular weight cut-off protein desalting column (Pierce - Thermofisher) pre-equilibrated in PBS to remove unbound dye and centrifuged for 30s at 3000g. The degree of protein labeling was estimated from measurement of absorbance at 280 nm and 650 nm (Dylight650) or 280nm and 493nm (Dylight488) was routinely between 2.5 and 3.5 of dye/mole of protein. Functionality of labeled Pros1 and Gas6 were confirmed by their capacity to augment TAM-dependent phagocytosis of apoptotic neutrophils (data not shown).

Flow cytometry analysis

All incubations were performed on ice to prevent internalization of antibody. Adherent MDM were detached by incubation in ice-cold HBSS without divalent cations (HBSS-/-) containing 0.5% (w/v) bovine serum albumin (BSA) and 2 mM EDTA for 15 min followed by vigorous pipetting. Detached cells were washed with HBSS -/- containing 2 mM EDTA, cells (3x10⁵-1.5x10⁶/assay) and resuspendend in 20 mM HEPES pH 7.4 containing 0.14 M NaCl and 0.1 % (w/v) BSA (flow buffer). Non-specific binding of antibodies to Fc receptors was reduced by pre-incubation for 5 min with 2% autologous serum. Cells were then labeled with saturating concentrations of fluorophore-conjugated antibodies for 30 min on ice. For assessment of binding of TAM ligands or Annexin V, cells were incubated with saturating concentrations of fluorescently labeled proteins diluted in 20 mM HEPES pH 7.4 containing 0.14 M NaCl with or without the addition of 2 mM CaCl₂ or 2.5 mM EDTA for 10 min on ice. Data was acquired using either a 5 laser LSR or 2 laser FACScalibur flow cytometer (BDBiosciences) and analysed with Flowjo software (Flowjo, Ashland, OR, USA).

Enzyme-linked immunosorbent assay (ELISA)

Cytokine release from 18h cultured monocytes following stimulation with 2 ng/mL LPS, 10 ng/mL IL-1ß or 50 ng/mL TNF was measured by ELISA. Monocyte culture media was removed and cells washed with HBSS without divalent cations containing 2 mM EDTA. Fresh IMDM was added containing 25 nM Pros1, Gla-less Pros1 or BMS 777607. After 20 min incubation in the presence or absence of Pros1, ΔGla Pros1 or inhibitor, LPS (2 ng/mL) was added and cells cultured for 8 h at 37°C, 5 % CO2. Cell culture supernatant was harvested and centrifuged at 6000g for 5 min to remove any cell remnants and stored at -20°C. ELISA for TNF, IL-6 IL-1ß and IL-8 were performed as indicated by manufacturer.

Statistics

Results are presented as mean ± s.d. or s.e.m. as indicated, where n = number of independent experiments using cells obtained from different donors and for multiple comparison tests, significance was analysed by ANOVA with Mann-Whitney post-test with Bonferroni correction for multiple comparisons using Instat software (www.graphpad.com).

Results

Expression of TAMs on monocytes

The expression of the TAM family of RTKs on human peripheral blood mononuclear cells was examined by flow cytometry. Monocytes were identified on the basis of forward/side scatter characteristics (Fig. 1A, upper panel) and expression of CD11b, CD11c, CD14, CD16, CD64 and HLA-DR (supplemental Fig. 1). We then used differential expression of CD14/CD16 (Fig. 1A, lower panel) to sub-divide the monocytes into CD14++/CD16- and CD14+/CD16+ populations(32) to determine the expression of Tyro3, Axl and Mer on these subsets (Fig. 1B). We did not detect expression of Axl on either of the monocyte populations when compared to binding of an isotype control antibody (Fig. 1B). In contrast, we consistently observed low levels of expression of Tyro-3 on monocytes (Fig. 1B). Mer expression was approximately 2-fold higher for the CD16+ expressing monocytes (Fig. 1B and quantified in Fig. 1C). To exclude the possibility that the expression of Mer we observed was mAb-dependent, a second anti-Mer antibody was used with similar results. In these experiments, Mer expression on CD16+ monocytes was found to be 1.7-fold higher than that of CD16 negative monocytes (n=3). Expression of Mer was higher for CD16+ monocytes further subdivided into high or intermediate expression when compared with CD16 negative monocytes (Mer on CD16 high: 1579±340; Mer on CD16 intermediate: 2047±202; Mer on CD16 negative: 862±85; IgG control: 280±100; mean fluorescence±s.d, n=3). Expression of TAMs was not a consequence of platelet binding to monocytes(33) as extensive washing of cells with EDTA to remove bound platelets did not affect Tyro3 or Mer expression. Freshly isolated human lymphocytes did not express Tyro3, Axl or Mer (data not shown).

Figure 1. Expression of TAMs on human monocyte subsets.

Levels of expression of TAM receptors on monocyte subsets present in human peripheral blood mononuclear cells were determined by labeling with a combination of APC-conjugated CD14 mAb/FITC-conjugated CD16 mAb and PE-conjugated isotype control, Axl, Tyro3, or Mer antibodies prior to flow cytometric analysis. A) Dot plots illustrating the gating strategy to define monocytes on the basis of laser scatter properties (upper panel) and CD14/CD16 antibody reactivity (lower panel). B) Histograms showing binding of antibodies specific for Axl, Tyro3 and Mer, compared with isotype control antibody (grey profile) on CD14++ (red profile) and CD14+/CD16+ (blue profile) monocyte subsets as identified in panel A. C) Quantification of mean fluorescence intensity of staining for antibodies on CD14++ (red bars) and CD14+/CD16+ (blue bars) monocyte subsets. Data shown are mean ± s.e.m., n=4. The difference in mean fluorescence for Mer antibody staining on CD14++ and CD14+/CD16+ monocytes was found to be significant by paired t-test analysis (p<0.01), whereas there was no significant difference for Tyro3 and Axl staining on monocyte subsets.

Binding of Pros1 to monocytes

We next investigated monocyte capacity to bind to TAM ligands. Since viable cells have been reported to expose low levels of PtdSer on their plasma membrane(24) which would potentially allow binding of TAM ligands, we examined binding of fluorescently labeled Pros1 or Gas6 in the presence or absence of Ca²⁺ to dissect PtdSer-dependent and PtdSer-independent components of binding. In addition, we compared binding of Annexin V to monocytes under the same assay conditions. Binding of Annexin V to monocytes was found to be strictly Ca²⁺-dependent, with no binding in the presence of EDTA (Fig. 2A). Importantly, Annexin V binding occurred independently of any morphological (nuclear condensation) or biochemical (activation of caspase 3) signs of monocyte apoptosis (supplemental Fig. 2 and data not shown). Furthermore, the levels of Annexin V binding to viable monocytes was 12.5 ± 3.3-fold lower than that for apoptotic monocytes, consistent with previous reports(24). These findings suggest that there is constitutive, low level exposure of PtdSer on the plasma membrane of viable monocytes.

Figure 2. Ca²⁺-dependent and Ca²⁺-independent binding of Protein S to monocytes.

Binding of Annexin V-APC, Protein S (Pros1) Dylight-650 or Gas6 Dylight-650 to monocytes (10 min at 4°C) was measured by flow cytometry. A) Representative histograms showing binding of labeled proteins to monocytes in the presence of 2mM CaCl₂ (red profile) or 2.5mM EDTA (blue profile) relative to that of labeled BSA (grey profile). B) Quantification of geometric mean fluorescence for the total binding of labeled proteins to monocytes in the presence of 2mM CaCl₂ (red bars) or 2.5mM EDTA (blue bars) was calculated using FlowJo and the mean ± s.e.m. for Annexin V (n=9), Pros1 (n=13), and Gas6 (n=5) is shown. C) Representative histograms showing flow cytometric analysis of binding of Pros1 and Pros1 lacking the Gla domain (ΔGla Pros1) to monocytes in the presence of 2mM CaCl₂ (red profile) or 2.5 mM EDTA (blue profile) relative to that of a control protein (BSA - grey profile).

Binding of both Pros1 and Gas6 (not shown) to monocytes was distinct from Annexin V binding in that there were Ca²⁺-dependent and -independent components (Fig. 2A, quantified in Fig. 2B). When we compared binding of plasma-derived Pros1 and Pros1 lacking the Gla domain (ΔGla Pros1), we observed binding to monocytes in the absence of Ca²⁺, consistent with a TAM-dependent binding component. In contrast, we did not observe Ca²⁺-dependent binding of ΔGla Pros1 (Fig. 2C). Together our data demonstrate that Pros1 binds to monocytes in both a PtdSer-dependent and –independent manner and that the PtdSer-dependent component of Pros1 binding requires the presence of the Gla-domain.

Binding of Pros1 to apoptotic neutrophils

Neutrophils undergo spontaneous apoptosis during in vitro culture(6) and a significant proportion of 18h cultured neutrophils expose PtdSer. Annexin V ((34) and data not shown) and Pros1 bound to a subset of neutrophils in the presence of Ca²⁺ but not in the presence of EDTA(28). These results contrast the data shown in Fig 2 for monocytes and suggest that Protein S only binds to apoptotic neutrophils via exposed PtdSer and that there was no Ca²⁺-independent binding. This suggestion was confirmed by demonstration of co-binding of Annexin V and Pros1, revealing that Annexin V positive cells also bound Pros1 (Fig. 3A), consistent with previous data for apoptotic thymocytes(35). Since binding of Pros1 to PtdSer is dependent on amino acid sequences present in the Gla domain, we predicted that ΔGla Pros1 would not bind to apoptotic neutrophils. As shown in Fig. 3B, we observed very low level binding of ΔGla Pros1 that was independent of the presence of Ca²⁺ (data not shown). Quantification of binding demonstrates that ΔGla Pros1 binds to apoptotic neutrophils at very low levels, consistent with the absence of the Ca²⁺-dependent PtdSer binding domain, (Fig. 3C). A small population of 18h cultured neutrophils were found to bind to both Pros1 and ΔGla Pros1 at high levels in either the presence or absence of Ca²⁺. These cells corresponded to propidium iodide positive necrotic cells (data not shown). Our data confirm high level opsonisation of apoptotic neutrophils with Pros1 but not ΔGla Pros1, consistent with binding to apoptotic neutrophils via interaction with PtdSer exposed on the plasma membrane.

Figure 3. Protein S-dependent monocyte phagocytosis of apoptotic neutrophils.

A) Neutrophils were cultured in vitro for 18h (to induce apoptosis), then were washed with EDTA prior to measurement of binding of fluorescently labeled Annexin V and Pros1 by flow cytometry. A) Representative histograms shows binding of Pros1 and Annexin V in combination in the presence of 5mM EDTA (Pros1 + Ann V + EDTA) to define Ca²⁺ independent binding as a control, binding of Pros1-Dylight488 (Pros1 only), binding of Annexin V-APC (Ann V only) alone or in combination (Pros1 + Ann V) in the presence of 2mM Ca²⁺. B) Binding of Pros1 and ΔGla Pros1 to 18h cultured neutrophils was assessed by flow cytometry. Neutrophils were washed with EDTA and representative histograms show the extent of binding of Dylight488-labeled Pros1 and Dylight488-labeled ΔGla Pros1 in the presence of 2mM Ca²⁺to the same preparation of apoptotic neutrophils. C) Quantification of the percentage of 18h cultured neutrophils that bind Dylight488-labeled Pros1 and ΔGla Pros1. The percentage of neutrophils that bound labeled proteins was measured in the presence of 2mM Ca²⁺ (black bars) or EDTA (gray bars). Data shown are mean ± s.e.m., n = 3. In the presence of 2mM Ca²⁺, the percentage of neutrophils binding Pros1 is significantly higher than ΔGla Pros1 (p<0.01), whereas there is no significant difference (NS) in percentage of neutrophils binding Pros1 and ΔGla Pros1 in the presence of EDTA. D) The capacity of monocytes for phagocytosis of apoptotic neutrophils was assessed using a flow cytometric assay. Cell Trace far red labeled 18h cultured monocytes were co-incubated with 18 h cultured autologous neutrophils that were labeled with pHrodo after washing in HBSS + 2.5mM EDTA. Monocytes and neutrophils were co-incubation at a cell ratio of 1:6 for 45 minutes in IMDM following pre-incubation of monocytes in IMDM (None) or 100 nM BMS777607 receptor tyrosine kinase inhibitor for 15 min (BMS), or with the addition of 25nM Pros1 (Pros1) or 25nM Pros1 following pre-incubation of monocytes in 100 nM BMS777607 for 15 min (Pros1 + BMS), or 25 nM ΔGla Pros1 (ΔGla). Representative histograms show the extent of pHrodo fluorescence for the far red labeled cells. E) Quantification of phagocytosis of apoptotic neutrophils by 18h monocytes in IMDM alone (untreated), monocytes pre-incubated with 100nM receptor tyrosine kinase inhibitor BMS777607 (BMS), monocytes with the addtion of 25nM Pros1, or monocytes pre-incubated with 100nM BMS777607 in the presence of 25nM Pros1 (Pros1 + BMS), or 25nM ΔGla Pros1. Data shown are mean ± s.e.m., n = 5. Statistical analysis using one-way ANOVA (Bonneferroni post-test) did not reveal a significant difference between any of the conditions marked (‡), whereas Pros1 (Δ) was found to be significantly different from all other conditions (p<0.005).

Phagocytosis of apoptotic neutrophils by 18h monocytes

We next compared 18h monocyte and monocyte-derived macrophage capacity for Pros1-dependent phagocytosis of apoptotic neutrophils. We used an experimental system in which monocytes and neutrophils were isolated from the same donor and cultured for 18h prior to use in an autologous cell interaction assay. Monocytes were isolated by negative immuno-magnetic selection to yield cells that were 89.3 ± 3.8% (n=8) CD64+ and were essentially free of bound platelets. Following in vitro culture for 18h, monocytes maintained expression of CD11b, CD162, Mer and Tyro3 (Supplemental Fig. 3A). Mer expression on 18 hour monocytes was further confirmed by immunoblot analysis (Supplemental Fig. 3C). Axl was not expressed by monocytes that had been cultured for 18h. In contrast to freshly isolated monocytes, expression of CD62L was down-regulated on 18h monocytes, consistent with initiation of monocyte-macrophage differentiation (Supplemental Fig. 3A). Since there was considerable overlap in laser scatter properties of cultured neutrophils and monocytes, we labeled monocytes with CellTrace far-red to allow discrimination from apoptotic neutrophil targets. Neutrophils were labeled with pHrodo to allow quantification of monocyte phagocytosis using flow cytometry. In the absence of Pros1 there was a small proportion of phagocytic monocytes (<5% phagocytosis; see example plots in Fig 3D, quantified in Fig 3E). Although phagocytosis was increased ~2-fold in the presence of Pros1, the absolute percentage (~6%) of monocytes capable of Pros1-dependent phagocytosis was low (Fig 3E). Pros1-dependent phagocytosis of apoptotic neutrophils by monocytes was blocked by pre-incubation with the receptor tyrosine kinase inhibitor, BMS777607, demonstrating a requirement for RTK activity(36). Consistent with the lack of binding to apoptotic neutrophils shown in Fig 3C, addition of Gla-less Pros1 failed to increase monocyte phagocytosis of apoptotic neutrophils (Fig 3D and quantified in Fig 3E), demonstrating a requirement for the PtdSer-binding domain of Pros1. Together, these data demonstrate that although 18h monocytes express both Mer and Tyro3, the capacity for Pros1-dependent phagocytosis of apoptotic cells is relatively low.

Phagocytosis of apoptotic neutrophils by monocyte-derived macrophages

To confirm that Pros1 was able to confer phagocytosis by monocyte-derived macrophages (MDM), we treated monocytes with glucocorticoids for 5 days to induce high levels of expression of Mer (36 and Supplemental Fig. 3D and E) and to promote Mer-dependent apoptotic cell phagocytosis. The proportion of glucocorticoid-treated (GC)-MDM capable of phagocytosis of pHrodo-labeled apoptotic neutrophils was ~10-fold higher than that of monocytes (baseline phagocytosis 44% versus 4.2% respectively; compare Fig 3E and Fig 4B). Phagocytosis of apoptotic neutrophils by GC-MDM was increased 1.9 fold in the presence of 25nM Pros1 (Fig. 4A – quantified in Fig. 4B). However, the absolute percentage increase in phagocytosis for GC-MDM in the presence of Pros1 was ~40%, with the majority of GC-MDM being phagocytic after 45 minutes. The Pros1-dependent increase in phagocytosis was not observed when GC-MDM were pre-incubated with the RTK inhibitor BMS777607, consistent with a requirement for Mer. In contrast, pre-incubation with BMS777607 did not affect GC-MDM phagocytosis of apoptotic neutrophils in the absence of Pros1. Finally, we tested our prediction that the ΔGla Pros1 would fail to confer increased phagocytosis by GC-MDM. When compared with full-length Pros1, equimolar concentrations of ΔGla Pros1 did not augment phagocytosis (Fig. 4A – quantified in Fig. 4B). Together, our data demonstrates that ΔGla Pros1 does not bind to apoptotic neutrophils and fails to act as a ligand for Mer-dependent phagocytosis of apoptotic cells by either 18h monocytes or GC-MDM.

Figure 4. Protein S Gla domain is necessary for Mer-dependent phagocytosis of apoptotic cells by monocyte-derived macrophages.

The capacity of dexamethsone-treated monocyte-derived macrophages (GC-MDM) for phagocytosis of apoptotic neutrophils was assessed using a flow cytometric assay. A) pHrodo-labeled 18 h cultured neutrophils were washed in HBSS + 2.5mM EDTA prior to co-incubation with 6d cultured MDM in 48 well plates at a ratio of 6:1 for 45 min in the absence of TAM ligands (None), or following pre-incubation of MDM with 100 nM BMS777607 receptor tyrosine kinase inhibitor for 15 min (None + BMS), or with 25nM Protein S (Pros1), or with 25nM Pros1 following pre-incubation with 100 nM BMS777607 for 15 min (Pros1 + BMS), or with 25 nM ΔGla Protein S (ΔGla Pros1). Representative flow cytometry histograms showing pHrodo fluorescence plotted against forward scatter for GC-MDM gated on their distinct forward/side scatter properties. B) Quantification of MDM phagocytosis of apoptotic neutrophils in the conditions described above for A). The percentage phagocytosis was calculated from analysis of the flow cytometry histograms using FlowJo and the mean ± s.e.m. is shown, (Pros1/None; n = 9, BMS/Pros1 + BMS; n=6, ΔGla Pros1; n=3). Statistical analysis using one-way ANOVA (Bonneferroni post-test) did not reveal a significant difference in the percentage phagocytosis observed under any of the conditions marked (‡), whereas Pros1 was found to be significantly increased when compared to absence of Pros1 (p<0.01).

Protein S augments LPS-dependent TNF production by monocytes

Although monocytes were inefficient phagocytes for apoptotic cells, they expressed both Mer and Tyro3 and were able to bind to Pros1. We therefore sought to examine the impact of Pros1 on monocyte pro-inflammatory cytokine production. Monocytes were co-cultured with apoptotic neutrophils in the presence or absence of the TLR4 ligand, LPS, for 8h. Consistent with published findings, we observed powerful suppression of TNF release in response to LPS following co-culture of monocytes with apoptotic neutrophils (Fig. 5A). Suppression of TNF release was observed whether Pros1 was present or not, indicating that the suppression was independent of Pros1 opsonization (Fig. 5A). We observed similar findings in experiments using MDM that had been cultured in IMDM and serum, although the effect was not as pronounced as for monocytes (Supplemental Fig. 4). However, since monocytes are not efficient phagocytes of apoptotic cells (see Fig. 3E), suppression of monocyte inflammatory cytokine production following co-culture with apoptotic cells was likely independent of phagocytosis(38). In the course of these experiments, we were surprised to find that when monocytes were incubated with LPS and Pros1 in the absence of apoptotic cells, there was a significant, 2-fold increase in TNF release. We confirmed that the Pros1 used in these experiments did not contain significant levels of contaminating endotoxin, with levels of LPS being <2pg/ml at the concentrations used in this assay. Furthermore, incubation of monocytes with Pros1 in the absence of LPS failed to induce TNF release (Fig 5A), suggesting our results were not a direct effect of Pros1 on monocyte TNF release. These data raised the possibility that engagement of monocyte TAMs in the absence of apoptotic cells acted to amplify toll-like receptor (TLR) responses.

Figure 5. Augmentation of monocyte TNF release by Protein S requires the Gla domain and Mer kinase activity.

A) Monocytes isolated by negative selection were cultured for 18h in IMDM containing 5% autologous serum, then washed in HBSS containing 2.5mM EDTA to remove bound Protein S prior to further cell culture. Monocytes were cultured in IMDM alone (control), or with 25nM Protein S (Pros1), or with LPS (2ng/ml), or with LPS and Pros1 together. In some experiments monocytes were cultured with LPS and autologous apoptotic neutrophils (neutrophil:monocyte ratio = 6:1) in the absence (AN) or presence of 25nM Protein S (AN + Pros1). After culture at 37°C for a further 8 h, cell culture supernatants were harvested and the levels of TNF measured by ELISA. Data shown are mean TNF release (ng/ml) ± s.e.m., n=8. Statistical analysis using one-way ANOVA (Bonneferroni post-test) revealed a significant difference in the levels of LPS-induced TNF release between control and Pros1 (p<0.01). B) LPS-induced TNF release from 18h monocytes was significantly increased by 75nM, 25nM and 7.5nM concentrations of Pros1. Data shown are mean TNF release (ng/ml) ± s.e.m., n=3, p<0.05. C) LPS-induced TNF release from 18h monocytes was significantly reduced following blockade of receptor tyrosine kinase activity by pre-incubation with 100nM BMS777607 (BMS) receptor tyrosine kinase inhibitor for 15 minutes prior to treatment with LPS/Pros1. Data shown are mean TNF release (ng/ml) ± s.e.m., n=6; p<0.05. D) 18h monocytes were cultured in IMDM alone, LPS alone (2ng/ml), LPS and 25nM Protein S, or 2ng/ml LPS and 25nM ΔGla Protein S together for 8 h. Cell supernatants were harvested and cytokine release was measured by ELISA. LPS-induced TNF release from monocytes was significantly increased in the presence of 25nM Pros1 but not in the presence of Pros1 lacking the Gla-domain (ΔGla Pros1). Data shown are mean TNF release (ng/ml) ± s.e.m., n=9; p<0.05. E) LPS-induced release of IL-6 from 18h monocytes was significantly increased in the presence of 25nM Pros1 but not in the presence of 25nM ΔGla Pros1. Data shown are mean IL-6 release (ng/ml) ± s.e.m., n=4; p<0.05.

The augmentation of LPS-induced TNF release from monocytes in the presence of Pros1 was concentration-dependent, with significant effects at 75, 25 and 7.5nM Pros1 (Fig. 5B). Furthermore, pre-treatment of monocytes with the RTK inhibitor BMS777607 to block TAM-dependent signaling prior to incubation with Pros1 and LPS inhibited the augmentation of TNF release, suggesting that TAM-dependent signaling was required (Fig. 5C). In contrast, ΔGla Pros1 failed to augment monocyte TNF production in response to LPS, indicating a dual requirement for the Gla-domain and PtdSer binding for augmentation of LPS-induced TNF release (Fig. 5D). Importantly, the effect of Pros1 was not restricted to release of TNF, as LPS-induced production of IL-6 was also increased in the presence of Pros1 (Fig. 5E). Together these data suggest that Pros1 has a synergistic pro-inflammatory effect upon monocyte cytokine production/release in response to TLR4 ligands.

Discussion

In this manuscript, we present important new findings relating to the TAM-mediated regulation of myeloid cell function:

First, human peripheral blood monocytes express Mer, low levels of Tyro3, but not Axl. Levels of expression of Mer were higher on CD14+/CD16+ expressing monocytes than CD14++ monocytes (Fig 1), consistent with increased expression associated with maturation status. However, we found that Mer-expressing monocytes were relatively poor phagocytes of Pros1-opsonised apoptotic cells (Fig 3). These data suggest that, for monocytes, additional receptors are required for the capture and internalization of apoptotic cells(39). Alternatively, there may be a threshold level of Mer or Tyro3 expression required to confer function as a phagocytic receptor(40).

Second, we observed low level binding of Annexin V to viable monocytes (as reported by others(24)) suggesting that PtdSer is exposed on the plasma membrane of viable monocytes. PtdSer is exposed on the surface of activated platelets where it has an important role in the regulation of coagulation(41). PtdSer exposure on viable cells is also critical for “pruning” of neuronal synapses by astrocytes and microglia(42) and removal of the outer segments of rod cells in the retina by retinal pigment epithelial cells(43). Exposure of PtdSer in viable myoblasts is critical for myotube formation(44). In addition, viable or activated leukocytes have also been shown to expose PtdSer(24, 45, 46), suggesting that PtdSer exposure is not restricted to apoptotic cells.

Using fluorescently labeled Pros1 and Gas6, we demonstrated that viable monocytes bind to TAM ligands (Fig 2) and that low level exposure of PtdSer on monocytes allows Ca²⁺-dependent binding of Pros1 that requires the presence of the Gla-domain. However, we also identified a Ca²⁺-independent component of Pros1 binding that does not require the Gla domain which we speculate is mediated via the sex hormone-binding globulin domain of Pros1. Ca²⁺-independent binding of TAM ligands was not observed for viable or apoptotic neutrophils (Fig 3A), or lymphocytes (data not shown).

Third, we observed Pros1-dependent augmentation of monocyte TNF release in response to LPS. This effect of Pros1 markedly contrasts the suppression of LPS-induced inflammatory cytokine release observed in the presence of apoptotic cells(9). One possibility is that Pros1 acted to alter monocyte viability or metabolism. However, we did not observe differential recovery of monocytes following incubation with Pros1 (82% versus 85%). Furthermore, metabolic profiling using Seahorse analysis did not reveal changes in aerobic or anaerobic energy utilization in Pros1-treated monocytes (data not shown). Data from preliminary experiments suggest that Pros1 also amplifies release of MCP1 and CXCL8 from LPS-treated monocytes (data not shown). In view of the lack of effect of Pros1 alone upon monocyte TNF release, we suggest that Pros1 has a synergistic effect upon LPS-mediated regulation of monocyte cytokine production/release. Augmentation of monocyte pro-inflammatory cytokine production by Pros1 was not seen with ΔGla Pros1, demonstrating a requirement for PtdSer binding for the observed augmentation of monocyte cytokine release.

Fourth, augmentation of LPS-induced TNF release by Pros1 was blocked by RTK inhibition, implying a requirement for TAM kinase activity. Previous studies have suggested that there are divergent signaling mechanisms downstream of Mer autophosphorylation, with activation of FAK/PLCγ being required for Rac activation necessary for phagocytosis(23). In contrast, Mer-dependent suppression of NFκB activation has been suggested to be independent of this pathway(23). Results from preliminary experiments showing that Pros1 does not affect pro-inflammatory cytokine release (TNF and CXCL8) in response to interleukin-1 (IL-1β) stimulation raise the possibility that there is a specific association between Mer and/or Tyro3 and TLR4-mediated signaling that leads to amplification of pro-inflammatory cytokine production. LPS has been reported to induce down-regulation of Mer, with loss of receptor expression ((47) and Supplemental Fig. 3B) and down-stream suppressive signaling potentially contributing to generation of pro-inflammatory cytokines(48). We therefore examined whether Pros1 affected the LPS-induced Mer downregulation at early timepoints. After 30 minutes of LPS treatment, expression of Mer was 61±12% of that at time=0 in the absence of Pros1 and 58±15% in the presence of 25nM Pros1. Similarly, after 60 minutes of LPS treatment, expression of Mer was 55±8% of that at time=0 in the absence of Pros1 and 56±11% in the presence of 25nM Pros1 (n=3). The reduction of expression of Mer induced by LPS was similar in the presence or absence of Pros1, suggesting that differential reduction in Mer expression did not account for increased pro-inflammatory cytokine production in the presence of Pros1.

Although it is well established that loss of Mer signaling affects pro-inflammatory cytokine release by macrophages(45), there may be species- or cell type-specific effects in addition to temporal differences in responses that may be important. We found similar effects of Pros1 upon LPS induced TNF release by human MDM, although we could not use the GC-MDM to further investigate the role of Mer as GC treatment of macrophages strongly down-regulates pro-inflammatory cytokine production. Thus, whilst treatment with GC drives Mer-dependent phagocytosis, examination of the potential role of Mer signaling in regulation of TNF production is not possible. It is likely that changes in Mer expression during human monocyte-macrophage differentiation(49) will also impact upon the engagement of downstream signaling pathways that drive pro- or anti-inflammatory signaling. Unlike human monocytes, mouse monocytes do not express Mer(47) and Pros1 dependent augmentation of TNF release may not be seen in mouse monocytes. Rothlin et al., demonstrated that for bone-marrow-derived dendritic cells, TAM-dependent suppression of LPS-induced IL-6 release was lost, or reversed in the absence of Signal transducer and activator of transcription-1 (STAT1) or IFNAR1 expression(21). In contrast, recent work from Earp and colleagues has identified a new molecular mechanism by which exogenous Pros1 acts via Mer and Tyro3 to suppress macrophage polarisation(51). In their studies, Mer formed a complex with protein tyrosine phosphatase 1B, acting to suppress intracellular signaling and pro-inflammatory gene expression. Interestingly, the suppressive effects of Pros1 upon some pro-inflammatory genes, including TNF and iNOS, were found to be less sensitive to Mer-dependent inhibition at early time points. The augmentation of TNF release by Pros1 we describe in this manuscript may account for the early augmentation of TNF release observed following short periods of co-culture of macrophages with apoptotic cells(52).

We suggest that TAMs could exert divergent regulatory effects upon monocyte/macrophage function during an inflammatory response in a manner that is dependent on the context of PtdSer exposure on the cell membrane (Figure 6). During the initiation of inflammatory responses, there will be few apoptotic cells present at inflammatory sites. We speculate that when ligation of Mer (and Tyro3) by TAM ligands is uncoupled from apoptotic cell phagocytosis, down-stream signaling may act to amplify, rather than suppress, LPS-induced monocyte pro-inflammatory mediator production. One possibility is that Pros1 binding to monocytes via PtdSer acts either in cis to engage TAMs on the same cell, or in trans to TAMs expressed on adjacent monocytes and confers these effects. In contrast, as inflammation progresses, the presence of apoptotic cells with high level opsonization by Pros1 would be predicted to promote TAM-dependent macrophage efferocytosis and engagement of anti-inflammatory signals that dampen pro-inflammatory cytokine release.

Figure 6. Schematic of possible mechanisms of Pros1 effects on LPS-induced TNF release by monocytes.

Exposure of PtdSer (dark grey circles) on the surface of apoptotic cells results in high level opsonisation with Pros1 (black rectangle). Pros1 induces maximal ligation of Mer RTK (grey oval), leading to suppression of LPS-induced signaling via CD14 and TLR4 (grey triangle) and inhibition of TNF release by monocytes. In contrast, low level binding of Pros1 to viable cells results in partial ligation of Mer RTK, initiating signals that act to amplify LPS-induced TNF release by monocytes.

In summary, our data suggests that in the absence of apoptotic cells, Pros1 acts to amplify LPS-dependent pro-inflammatory cytokine production by monocytes, an effect that requires RTK activity and the presence of the PtdSer-binding Gla domain of Pros1. Our findings unveil potential new roles for TAMs in regulation of monocyte/macrophage function during progression of inflammation.